Generative AI Web App using Python Flask with Amazon Bedrock

Let’s create a generative AI-enabled web application from scratch using Python Flask. The application will provide a cloud security tip of the day by integrating with Amazon Bedrock using the amazon.titan-text-premier-v1:0 model.

This blog is an introduction to building a generative AI web application using Amazon Bedrock. The application provides a daily cloud security tip, however please note that detailed security aspects about Amazon Bedrock will be covered in the future. Stay tuned for more updates!

Prerequisites:

An AWS account with access to Amazon Bedrock with `amazon.titan-text-premier-v1:0` foundation model enabled.
Python installed on your system.
Basic knowledge of Flask.

Steps:

Create a new project directory:

mkdir cloud-security-tips
cd cloud-security-tips

Set up a virtual environment:

python3 -m venv venv
source venv/bin/activate  # On Windows use `venv\Scripts\activate`

Install Flask and Boto3 (AWS SDK for Python):

python3 -m pip install flask boto3

Create a filename app.pyto create python3 flask app:

Use AWS secret manager or open source secret managers instead of hardcoded credentials.

touch app.py

from flask import Flask, render_template, request, jsonify
import boto3
import os

app = Flask(__name__)

# Load AWS credentials from environment variables
AWS_ACCESS_KEY = "AKIA2SAMPLE" #changethis
AWS_SECRET_KEY = "0nu-SAMPLEKEY-rJpP" #changethis
AWS_REGION = 'us-east-1'  # Change to your region


# Initialize the Amazon Bedrock client
bedrock_client = boto3.client('bedrock-runtime',
                              aws_access_key_id=AWS_ACCESS_KEY,
                              aws_secret_access_key=AWS_SECRET_KEY,
                              region_name=AWS_REGION)

# Home route
@app.route('/')
def index():
    return render_template('index.html')

# API route to get cloud security tip
@app.route('/get_tip', methods=['GET'])
def get_tip():
    model_id = 'amazon.titan-text-premier-v1:0'
    prompt = "Provide a cloud security tip of the day"
    response = bedrock_client.converse(
        modelId=model_id,
        messages=[{"role": "user", "content": [{"text": prompt}]}]
    )
    tip = response['output']['message']['content'][0]['text']
    return jsonify({'tip': tip})

if __name__ == '__main__':
    app.run(debug=True)

Create a templates directory and an index.html file inside it:

mkdir templates
touch index.html

<!DOCTYPE html>
<html lang="en">
<head>
    <meta charset="UTF-8">
    <meta name="viewport" content="width=device-width, initial-scale=1.0">
    <title>Cloud Security Tips</title>
    <style>
        body { font-family: Arial, sans-serif; background-color: #f0f0f0; text-align: center; padding: 50px; }
        #tip { font-size: 1.5em; margin-top: 20px; }
        button { padding: 10px 20px; font-size: 1em; }
    </style>
</head>
<body>
    <h1>Cloud Security Tip of the Day</h1>
    <button onclick="getTip()">Get Tip</button>
    <div id="tip">Click the button to get today's cloud security tip!</div>
    <script>
        async function getTip() {
            const response = await fetch('/get_tip');
            const data = await response.json();
            document.getElementById('tip').innerText = data.tip;
        }
    </script>
</body>
</html>

Run the Flask application:

python3 app.py

Open your browser and navigate to http://127.0.0.1:5000. You should see the web page with a button to get the cloud security tip of the day.

Source code: https://github.com/justmorpheus

Explanation

app.py:
– Imports necessary modules for setting up flask.
– Setup credentials (for testing), in production/reality do not hard-code credentials.

Conclusion

By following these steps, you can build and run a generative AI web application using Python Flask and Amazon Bedrock.

In case of issues associated with converse method not available, refer to github issues.

References & Credits

Continue reading on Medium

Enjoyed this article? Visit Medium to leave a comment, clap, or follow Divyanshu for more insights!

Read on Medium

Prerequisites:

Source code: https://github.com/justmorpheus

Explanation

Conclusion

References & Credits

Continue reading on Medium

More Articles

How I Found a Path Traversal in the Rancher Dashboard via HAR Replay

EKS Moved To containerd, GuardDuty’s Runtime Alert Stayed Behind. Here’s Why That Matters.